[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::fddi

Title:	FDDI - The Next Generation

Moderator:	NETCAD::STEFANI

Created:	Thu Apr 27 1989
Last Modified:	Thu Jun 05 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	2259
Total number of notes:	8590

96.0. "Basic confusion on FDDI mechanism" by KETJE::DEMOOR () Fri Jul 27 1990 04:07

    While reading Raj Jain's report (DEC-TR-655) on performance analysis,
    something got me confused about understanding FDDI basics.
    
    On page 3, Figure 1 explains a typical chronology of events in frame
    transmission.
    
    One can understand that the next station starts receiving
    the first bits of a frame while the sending station is still busy
    putting the rest of the frame on the cable. So, I can accept that the
    sending of the frame by station N (t4 -> t7) and receiving the frame by
    station N+1 (t5 -> t8) overlap. I suppose (t5 - t4) is propagation
    delay.
    
    What gets me confused is "t6 The first bit of the frame arrives back at
    the source station, wich begin to strip it".
    
    What makes the source station say "Ha ! here's MY first bit ! " ?
    Doesn't it have to wait until it figures out the sending addresses ?
    Will it not wait untill full reception of the frame to check things out
    ? FCS covers also DA and SA, so FCS can tell you if the SA is still
    valid.
    
    I GUESS that the actual action of "stripping" and "copying" is in fact:
    - the whole frame has always been stored in a buffer anyway (for FCS) ?
    - Copying is an action that is performed AFTER reception and BEFORE or
      WHILE retransmission to next station ?
    - Stripping is a decision point in the algorithm, where it is decided
      to transmit a token frame, rather than the one just received.
    
    In all drawings on DVN video, Electronic Seminar handout, etc, all
    these action are represented in a (symbolic) way that makes one believe
    that a station starts retransmitting a frame while it was not yet fully
    received.
    
    I personally understand things better if the description is accurate.
    Can somebody knowledgeable clarify ?
    
    Regards,
    
    Dominique.

T.R	Title	User	Personal Name	Date	Lines
96.1		KONING::KONING	NI1D @FN42eq	`Fri Jul 27 1990 13:08`	19
	Stripping starts immediately after the SA. Apart from bridges (where the answer is rather complicated) stripping is based on a simple comparison of addresses. If the frame has been corrupted, and the SA was part of what changed, then this may cause the "wrong" node to strip the frame, but that's not a problem since no one wants the frame anyway. So you need more than the first bit before stripping starts, but, fortunately, far less than a whole frame. Note that the nodes repeat the frame as they look for the SA. So when the decision to strip occurs, the first 13 bytes of the frame (FC, DA, SA) have already been repeated. This works, because the stripping turns the frame into a "fragment": the SA is immediately followed by Idle symbols. Since there isn't a "Terminator" symbol after the data, a fragment is NOT a frame and is totally ignored by everyone. Fragments keep circulating until some node transmits, at which point they disappear. In particular, they do not "use up bandwidth". paul
96.2	Forwarding & stripping basics	KETJE::DEMOOR		`Mon Jul 30 1990 04:01`	27
	Paul, >> Note that the nodes repeat the frame as they look for the SA. So, my understanding of FDDI basics was indeed completely wrong !! A station repeats the frame while receiving ! Does it repeat the bits, or does it receive and regenerate the symbols ? Are there any considerations here regarding increasing bit error rate as the frame gets forwarded ? >> ....then this may cause the "wrong" node to strip the frame, >> but that's not a problem since no one wants the frame anyway. Well, I would mind if this node is sitting BEFORE me, and strips a frame that had me as destination ! >> Fragments keep circulating until some node transmits, at which point they >> disappear. In particular, they do not "use up bandwidth". I don't see how this "disappearing" mechanism works. Regular frames disappear by stripping. But what keeps the stations from forwarding these "fragments" forever ? From your explanantion, I just understood that the fragment bytes are forwarded without discernation. Regards, Dominique.
96.3		KONING::KONING	NI1D @FN42eq	`Mon Jul 30 1990 11:14`	15
	A station forwards the symbols as it repeats. In the process it checks the frame, and flags it as bad (in the "E indicator" at the end of the frame) if it sees that the CRC is bad. If a frame is sent off to you, and then the SA is mangled so it looks like a different node sent it, the CRC will be bad. You don't look at frames with bad CRC, do you? So if the frame is stripped before it gets to you that doesn't matter. When a station wants to transmit, it grabs the token and starts to transmit. If there are any fragments on the ring, they will be arriving at that station's receiver in the meantime. Stations don't repeat while transmitting, so that's how fragments go away. paul